Corpus-Based Error Analysis of Students’ Online Writings in Automated Essay Scoring System

Abstract

Based on the corpora of 60 pieces of students’ writings by the Contestant Candidates for National English Writing Contest for Technical and Vocational Students in recent years, the paper makes a comparative error analysis of expository writings by English Majors and Non-English Majors in a vocational college in Shanghai, utilizing tools such as PatCount, AntConc, and SPSS 19. Three main findings are arrived at as follows: first, there are significant differences between the expository writing by English major contestants and by their Non-English major counterparts, especially in the writing quality, where the former is obviously superior to the latter, despite that there were not apparent differences in their English proficiencies in the college entrance examination. Second, the non-English major contestants commit no errors in pronoun, which does not necessarily mean they are better at using pronoun than the English major contestants. Third, the most frequently committed errors by both the English majors and the non-English majors are those errors in sentence, wording, and word form. And the main causes for the occurrences of those errors are two factors, that is, the negative transfer of the mother tongue, and the carelessness of the learners. Finally, some pedagogical suggestions are made on how to improve the writing quality of the students.